Hugging Face's logo

admko
/
sembr2023-bert-tiny

Token Classification
Transformers
PyTorch
bert
Generated from Trainer
Inference Endpoints
Model card Files Community
sembr2023-bert-tiny / README.md
admko's picture
admko
End of training
e040f43
preview code
|
raw
history blame
16.1 kB
metadata
license: mit
base_model: prajjwal1/bert-tiny
tags:
  - generated_from_trainer
datasets:
  - sembr2023
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: sembr2023-bert-tiny
    results:
      - task:
          name: Token Classification
          type: token-classification
        dataset:
          name: sembr2023
          type: sembr2023
          config: sembr2023
          split: test
          args: sembr2023
        metrics:
          - name: Precision
            type: precision
            value: 0.7287362872204017
          - name: Recall
            type: recall
            value: 0.6756042794875181
          - name: F1
            type: f1
            value: 0.7011651816312543
          - name: Accuracy
            type: accuracy
            value: 0.9490469679439985

sembr2023-bert-tiny

This model is a fine-tuned version of prajjwal1/bert-tiny on the sembr2023 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2172
  • Precision: 0.7287
  • Recall: 0.6756
  • F1: 0.7012
  • Iou: 0.5398
  • Accuracy: 0.9490
  • Balanced Accuracy: 0.8256
  • Overall Accuracy: 0.9348

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 0.0001
  • train_batch_size: 64
  • eval_batch_size: 128
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • training_steps: 1000

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Iou Accuracy Balanced Accuracy Overall Accuracy
1.422 0.07 10 1.3253 0.0 0.0 0.0 0.0 0.9114 0.4999 0.9114
0.8436 0.14 20 0.7947 0 0.0 0.0 0.0 0.9115 0.5 0.9115
0.6272 0.21 30 0.5924 0 0.0 0.0 0.0 0.9115 0.5 0.9115
0.4967 0.28 40 0.5178 0 0.0 0.0 0.0 0.9115 0.5 0.9115
0.4864 0.35 50 0.4818 0 0.0 0.0 0.0 0.9115 0.5 0.9115
0.4554 0.42 60 0.4575 0 0.0 0.0 0.0 0.9115 0.5 0.9115
0.4635 0.49 70 0.4410 0 0.0 0.0 0.0 0.9115 0.5 0.9115
0.4238 0.56 80 0.4261 0 0.0 0.0 0.0 0.9115 0.5 0.9115
0.4104 0.62 90 0.4136 0 0.0 0.0 0.0 0.9115 0.5 0.9115
0.3832 0.69 100 0.3958 0 0.0 0.0 0.0 0.9115 0.5 0.9115
0.3664 0.76 110 0.3693 0 0.0 0.0 0.0 0.9115 0.5 0.9115
0.3736 0.83 120 0.3477 0.5263 0.0013 0.0026 0.0013 0.9115 0.5006 0.9115
0.3364 0.9 130 0.3317 0.5220 0.0110 0.0215 0.0109 0.9116 0.5050 0.9115
0.3095 0.97 140 0.3221 0.5584 0.0556 0.1011 0.0533 0.9126 0.5257 0.9118
0.3233 1.04 150 0.3156 0.5713 0.1471 0.2340 0.1325 0.9148 0.5682 0.9129
0.3072 1.11 160 0.3112 0.6452 0.1640 0.2616 0.1505 0.9181 0.5776 0.9160
0.287 1.18 170 0.3075 0.6336 0.2622 0.3709 0.2277 0.9213 0.6237 0.9177
0.3061 1.25 180 0.3039 0.7025 0.2142 0.3283 0.1964 0.9224 0.6027 0.9189
0.2924 1.32 190 0.2997 0.7581 0.1511 0.2520 0.1442 0.9206 0.5732 0.9183
0.3081 1.39 200 0.2950 0.7367 0.2025 0.3177 0.1888 0.9230 0.5977 0.9193
0.2993 1.46 210 0.2921 0.6861 0.2902 0.4079 0.2562 0.9255 0.6387 0.9200
0.275 1.53 220 0.2885 0.6734 0.3249 0.4383 0.2807 0.9263 0.6548 0.9200
0.2692 1.6 230 0.2861 0.6622 0.3438 0.4526 0.2925 0.9264 0.6634 0.9188
0.2536 1.67 240 0.2828 0.6295 0.3895 0.4812 0.3169 0.9257 0.6836 0.9176
0.265 1.74 250 0.2790 0.6586 0.3546 0.4610 0.2996 0.9266 0.6684 0.9200
0.2571 1.81 260 0.2736 0.6641 0.3729 0.4776 0.3137 0.9278 0.6773 0.9208
0.2684 1.88 270 0.2711 0.6794 0.4142 0.5146 0.3465 0.9309 0.6976 0.9222
0.2754 1.94 280 0.2721 0.6408 0.4813 0.5497 0.3790 0.9302 0.7276 0.9198
0.2507 2.01 290 0.2652 0.6922 0.4438 0.5408 0.3707 0.9333 0.7123 0.9240
0.2678 2.08 300 0.2622 0.6957 0.4243 0.5271 0.3578 0.9326 0.7031 0.9244
0.2676 2.15 310 0.2629 0.6698 0.5020 0.5739 0.4024 0.9340 0.7390 0.9228
0.2369 2.22 320 0.2596 0.6667 0.5123 0.5794 0.4079 0.9342 0.7437 0.9233
0.2293 2.29 330 0.2560 0.6741 0.5174 0.5854 0.4138 0.9352 0.7465 0.9240
0.235 2.36 340 0.2517 0.7160 0.4796 0.5744 0.4030 0.9371 0.7306 0.9270
0.209 2.43 350 0.2545 0.6704 0.5332 0.5940 0.4225 0.9355 0.7539 0.9236
0.2032 2.5 360 0.2486 0.7002 0.5130 0.5922 0.4206 0.9375 0.7458 0.9267
0.2005 2.57 370 0.2482 0.6870 0.5418 0.6058 0.4345 0.9376 0.7589 0.9259
0.206 2.64 380 0.2463 0.6949 0.5384 0.6067 0.4354 0.9382 0.7577 0.9264
0.2196 2.71 390 0.2421 0.7158 0.5237 0.6049 0.4336 0.9395 0.7518 0.9289
0.1863 2.78 400 0.2410 0.7072 0.5435 0.6146 0.4437 0.9397 0.7608 0.9282
0.2036 2.85 410 0.2408 0.6889 0.5775 0.6283 0.4580 0.9395 0.7761 0.9272
0.1982 2.92 420 0.2345 0.7198 0.5596 0.6297 0.4595 0.9418 0.7692 0.9300
0.1909 2.99 430 0.2327 0.7173 0.5830 0.6432 0.4741 0.9428 0.7804 0.9307
0.2286 3.06 440 0.2309 0.7250 0.5933 0.6526 0.4843 0.9441 0.7857 0.9317
0.1839 3.12 450 0.2305 0.7219 0.6159 0.6647 0.4978 0.9450 0.7964 0.9323
0.2086 3.19 460 0.2280 0.7252 0.6171 0.6668 0.5002 0.9454 0.7972 0.9328
0.2055 3.26 470 0.2302 0.7088 0.6450 0.6754 0.5099 0.9451 0.8096 0.9318
0.1925 3.33 480 0.2252 0.7309 0.6263 0.6746 0.5090 0.9465 0.8020 0.9336
0.165 3.4 490 0.2248 0.7254 0.6364 0.6780 0.5128 0.9465 0.8065 0.9336
0.1814 3.47 500 0.2283 0.7008 0.6637 0.6818 0.5172 0.9452 0.8181 0.9314
0.1812 3.54 510 0.2239 0.7275 0.6436 0.6830 0.5186 0.9471 0.8101 0.9336
0.1738 3.61 520 0.2237 0.7241 0.6498 0.6850 0.5209 0.9471 0.8129 0.9335
0.1726 3.68 530 0.2227 0.7271 0.6517 0.6873 0.5236 0.9475 0.8140 0.9338
0.188 3.75 540 0.2204 0.7407 0.6393 0.6863 0.5224 0.9483 0.8088 0.9348
0.187 3.82 550 0.2185 0.7539 0.6303 0.6866 0.5227 0.9491 0.8052 0.9362
0.1917 3.89 560 0.2193 0.7354 0.6532 0.6919 0.5289 0.9485 0.8152 0.9349
0.1794 3.96 570 0.2197 0.7326 0.6574 0.6929 0.5301 0.9485 0.8170 0.9346
0.1541 4.03 580 0.2203 0.7292 0.6645 0.6954 0.5330 0.9485 0.8203 0.9343
0.1837 4.1 590 0.2190 0.7336 0.6563 0.6928 0.5300 0.9485 0.8166 0.9349
0.1541 4.17 600 0.2177 0.7405 0.6467 0.6904 0.5272 0.9487 0.8123 0.9354
0.1721 4.24 610 0.2210 0.7178 0.6767 0.6966 0.5345 0.9479 0.8254 0.9338
0.1657 4.31 620 0.2186 0.7323 0.6628 0.6958 0.5335 0.9487 0.8196 0.9350
0.1792 4.38 630 0.2182 0.7294 0.6668 0.6967 0.5345 0.9486 0.8214 0.9349
0.1908 4.44 640 0.2183 0.7309 0.6648 0.6963 0.5341 0.9487 0.8205 0.9348
0.1581 4.51 650 0.2177 0.7330 0.6658 0.6978 0.5359 0.9490 0.8211 0.9349
0.169 4.58 660 0.2178 0.7313 0.6685 0.6985 0.5366 0.9489 0.8223 0.9347
0.1756 4.65 670 0.2184 0.7271 0.6723 0.6986 0.5369 0.9487 0.8239 0.9344
0.1563 4.72 680 0.2179 0.7311 0.6706 0.6996 0.5379 0.9490 0.8233 0.9349
0.1684 4.79 690 0.2161 0.7475 0.6565 0.6990 0.5373 0.9500 0.8175 0.9362
0.1585 4.86 700 0.2171 0.7380 0.6648 0.6995 0.5378 0.9495 0.8209 0.9354
0.1841 4.93 710 0.2181 0.7283 0.6745 0.7004 0.5389 0.9489 0.8251 0.9346
0.1724 5.0 720 0.2177 0.7323 0.6695 0.6995 0.5379 0.9491 0.8229 0.9349
0.1791 5.07 730 0.2170 0.7329 0.6708 0.7005 0.5391 0.9492 0.8236 0.9351
0.1712 5.14 740 0.2171 0.7344 0.6705 0.7010 0.5396 0.9494 0.8235 0.9354
0.1489 5.21 750 0.2164 0.7374 0.6683 0.7012 0.5398 0.9496 0.8226 0.9357
0.157 5.28 760 0.2161 0.7407 0.6636 0.7000 0.5385 0.9497 0.8205 0.9358
0.1686 5.35 770 0.2180 0.7262 0.6775 0.7010 0.5396 0.9489 0.8263 0.9345
0.1526 5.42 780 0.2168 0.7344 0.6690 0.7002 0.5387 0.9493 0.8228 0.9351
0.1542 5.49 790 0.2172 0.7313 0.6722 0.7005 0.5390 0.9491 0.8241 0.9349
0.1498 5.56 800 0.2168 0.7351 0.6691 0.7005 0.5391 0.9494 0.8229 0.9353
0.1571 5.62 810 0.2167 0.7348 0.6687 0.7002 0.5387 0.9493 0.8227 0.9354
0.1682 5.69 820 0.2175 0.7265 0.6775 0.7011 0.5398 0.9489 0.8263 0.9347
0.1688 5.76 830 0.2175 0.7267 0.6764 0.7006 0.5392 0.9489 0.8259 0.9347
0.1622 5.83 840 0.2161 0.7393 0.6633 0.6992 0.5376 0.9495 0.8203 0.9357
0.1547 5.9 850 0.2173 0.7282 0.6755 0.7008 0.5395 0.9490 0.8255 0.9347
0.1712 5.97 860 0.2166 0.7339 0.6701 0.7005 0.5391 0.9493 0.8232 0.9352
0.1632 6.04 870 0.2168 0.7317 0.6724 0.7008 0.5394 0.9492 0.8242 0.9352
0.1615 6.11 880 0.2167 0.7315 0.6727 0.7009 0.5395 0.9492 0.8244 0.9352
0.1543 6.18 890 0.2164 0.7348 0.6699 0.7008 0.5395 0.9494 0.8232 0.9354
0.1407 6.25 900 0.2168 0.7318 0.6726 0.7009 0.5396 0.9492 0.8243 0.9351
0.1607 6.32 910 0.2170 0.7299 0.6743 0.7010 0.5396 0.9491 0.8250 0.9350
0.1652 6.39 920 0.2172 0.7276 0.6760 0.7009 0.5395 0.9489 0.8257 0.9347
0.1676 6.46 930 0.2173 0.7274 0.6765 0.7010 0.5397 0.9489 0.8260 0.9347
0.14 6.53 940 0.2174 0.7267 0.6767 0.7008 0.5394 0.9489 0.8260 0.9346
0.1634 6.6 950 0.2173 0.7276 0.6764 0.7011 0.5397 0.9490 0.8259 0.9347
0.174 6.67 960 0.2172 0.7283 0.6759 0.7011 0.5398 0.9490 0.8257 0.9348
0.156 6.74 970 0.2172 0.7287 0.6759 0.7013 0.5400 0.9491 0.8257 0.9348
0.1641 6.81 980 0.2172 0.7287 0.6756 0.7012 0.5398 0.9490 0.8256 0.9348
0.1634 6.88 990 0.2172 0.7287 0.6756 0.7012 0.5398 0.9490 0.8256 0.9348
0.1753 6.94 1000 0.2172 0.7287 0.6756 0.7012 0.5398 0.9490 0.8256 0.9348

Framework versions

  • Transformers 4.34.1
  • Pytorch 2.0.1
  • Datasets 2.14.6
  • Tokenizers 0.14.1